UTF8 patch for Subversion 1.8.0

utf8_patch.diff

--- subversion/libsvn_subr/path.c.original	2013-07-02 23:04:31.000000000 +0200
+++ subversion/libsvn_subr/path.c	2013-07-02 23:29:30.000000000 +0200
@@ -40,6 +40,9 @@
 
 #include "dirent_uri.h"
 
+#if defined(DARWIN)
+#include <CoreFoundation/CoreFoundation.h>
+#endif /* DARWIN */
 
 /* The canonical empty path.  Can this be changed?  Well, change the empty
    test below and the path library will work, not so sure about the fs/wc
@@ -1100,7 +1103,6 @@
 }
 
 
-#if !defined(WIN32) && !defined(DARWIN)
 /** Get APR's internal path encoding. */
 static svn_error_t *
 get_path_encoding(svn_boolean_t *path_is_utf8, apr_pool_t *pool)
@@ -1119,7 +1121,6 @@
   *path_is_utf8 = (encoding_style == APR_FILEPATH_ENCODING_UTF8);
   return SVN_NO_ERROR;
 }
-#endif
 
 
 svn_error_t *
@@ -1127,19 +1128,15 @@
                            const char *path_utf8,
                            apr_pool_t *pool)
 {
-#if !defined(WIN32) && !defined(DARWIN)
   svn_boolean_t path_is_utf8;
   SVN_ERR(get_path_encoding(&path_is_utf8, pool));
   if (path_is_utf8)
-#endif
     {
       *path_apr = apr_pstrdup(pool, path_utf8);
       return SVN_NO_ERROR;
     }
-#if !defined(WIN32) && !defined(DARWIN)
   else
     return svn_utf_cstring_from_utf8(path_apr, path_utf8, pool);
-#endif
 }
 
 
@@ -1148,16 +1145,43 @@
                          const char *path_apr,
                          apr_pool_t *pool)
 {
-#if !defined(WIN32) && !defined(DARWIN)
   svn_boolean_t path_is_utf8;
+#if defined(DARWIN)
+  svn_error_t *err;
+  /*
+   Compose any decomposed unicode characters precomposed one.
+   This will solve the problem that the 'svn status' command sometime
+   cannot recognize as same file when files suppose to be comtain
+   comopsed characters, like umlaut in some European language or
+   'Daku-ten' in Japanese, and the files are added on windows machines
+   then you use svn on Mac OS X checking out the files.
+   */
+  CFMutableStringRef cfmsr = CFStringCreateMutable(NULL, 0);
+  CFStringAppendCString(cfmsr, path_apr, kCFStringEncodingUTF8);
+  CFStringNormalize(cfmsr, kCFStringNormalizationFormC);
+  CFIndex path_buff_size = 1 + CFStringGetMaximumSizeForEncoding(
+                                                                 CFStringGetLength(cfmsr), kCFStringEncodingUTF8);
+  char *path = apr_palloc(pool, path_buff_size);
+  CFStringGetCString(cfmsr, path, path_buff_size, kCFStringEncodingUTF8);
+
   SVN_ERR(get_path_encoding(&path_is_utf8, pool));
+
   if (path_is_utf8)
-#endif
-    {
-      *path_utf8 = apr_pstrdup(pool, path_apr);
-      return SVN_NO_ERROR;
-    }
-#if !defined(WIN32) && !defined(DARWIN)
+  {
+    *path_utf8 = apr_pstrdup(pool, path);
+    err = SVN_NO_ERROR;
+  }
+  else
+    err = svn_utf_cstring_to_utf8(path_utf8, path, pool);
+  CFRelease(cfmsr);
+  return err;
+#else
+  SVN_ERR(get_path_encoding(&path_is_utf8, pool));
+  if (path_is_utf8)
+  {
+    *path_utf8 = apr_pstrdup(pool, path_apr);
+    return SVN_NO_ERROR;
+  }
   else
     return svn_utf_cstring_to_utf8(path_utf8, path_apr, pool);
 #endif

Does apply and svn stats runs with no direct error but does not fix the utf8 issue.